Load forecasting in electricity distribution
Hourly consumption forecast error fell 28%, reducing balancing costs.
- Sector
- Energy
- Product used
- Ilmek Tahmin
- Measured impact
- Forecast error -28%
- Time to production
- 8
High-risk subscribers are flagged 45 days ahead and given personalised offers, cutting churn by 18%.
A telecom operator ran blanket discount campaigns to reduce churn. Ilmek Öneri calculates churn probability from usage, billing and complaint history and recommends the best offer and channel for each subscriber.
Compared with the control group monthly subscriber loss fell 18%; campaign cost fell 30% because offers went only to relevant subscribers.
| Monthly churn | -%18 |
|---|---|
| Campaign cost | -%30 |
Hourly consumption forecast error fell 28%, reducing balancing costs.
62% of order, return and shipping questions are resolved without human intervention.
All calls are transcribed and checked for mandatory statements; quality team workload fell by 60%.